The other day, while I was doing a bit of research into the Riemann Zeta function, I discovered a few interesting closed-form values of summations involving it, like $$\sum_{s=2}^\infty (\zeta(s)-1)=1$$ $$\sum_{s=1}^\infty (\zeta(2s)-1)=\frac{3}{4}$$ and when I checked Wikipedia, I found a ton of other more complicated sums involving the Zeta Function, like $$\sum_{s=2}^\infty \frac{\zeta(s)-1}{s}=1-\gamma$$ I know the trick that can be used for these sums is switching the order of the sums and using the fact that a geometric series is formed, but it fascinated me nevertheless.

QUESTION: Does anybody know of any other interesting examples of sums (that don't telescope trivially) with each term using non-elementary functions but a final answer that is elementary, or expressible in terms of other widely-known mathematical constants (like $\pi$, $e$, or $\gamma$)?

Specifically, I am looking for interesting identities regarding:

  • sums involving the inverses of functions that don't have elementary inverses, like the Lambert-W function, or the inverse of the function $f(x)=x+\sin x$
  • sums involving $\Gamma(s)$, at non-integer values of $s$
  • interesting generating functions for sequences that currently have no discovered closed-form explicit formula, like the Bell, Bernoulli, or Harmonic numbers
  • sums involving $\operatorname{Si}(s)$, $\operatorname{Li}(s)$, or $\operatorname{Ei}(s)$, or the elliptic integrals


  • 1
    $\begingroup$ Maybe also you are interested in Gregory coefficients, see this Wikipedia. I believe that Gregory was contemporary with Newton. Good luck. $\endgroup$ – user243301 Sep 11 '17 at 15:42
  • 1
    $\begingroup$ @user243301 : Good post, I think it fits well to the topic. :-) $\endgroup$ – user90369 Sep 11 '17 at 16:25
  • 1
    $\begingroup$ @user243301 Interesting, thanks! $\endgroup$ – Frpzzd Sep 12 '17 at 23:00

In the non-exhaustive list below, you can find a lot of series involving special functions.

Among them many are interesting. For example, this one involving $\Gamma$ and $\zeta$ : $$\sum_{k=1}^\infty \frac{(-1)^k}{k!}\Gamma(k+x)\zeta(k+x)=-\Gamma(x)$$ Especially with $x=\frac{1}{2}$ one get a series representation for $-\sqrt{\pi}$.

It should be too long to edit many others from :

http://functions.wolfram.com/ZetaFunctionsandPolylogarithms/Zeta/23/01/ http://functions.wolfram.com/ZetaFunctionsandPolylogarithms/PolyLog/23/01/ http://functions.wolfram.com/Polynomials/HermiteH/23/02/ http://functions.wolfram.com/Polynomials/LaguerreL/23/02/ http://functions.wolfram.com/Polynomials/LegendreP/23/02/ http://functions.wolfram.com/Polynomials/ChebyshevT/23/01/ http://functions.wolfram.com/Polynomials/ChebyshevU/23/01/ http://functions.wolfram.com/Polynomials/GegenbauerC3/23/02/ http://functions.wolfram.com/Polynomials/JacobiP/23/01/ http://functions.wolfram.com/Polynomials/EulerE2/23/02/ http://functions.wolfram.com/Polynomials/BernoulliB2/23/02/ http://functions.wolfram.com/Polynomials/BellB2/23/02/ http://functions.wolfram.com/Polynomials/NorlundB2/23/02/ http://functions.wolfram.com/Polynomials/ZernikeR/23/01/ http://functions.wolfram.com/HypergeometricFunctions/ChebyshevTGeneral/23/01/ http://functions.wolfram.com/HypergeometricFunctions/GegenbauerC/23/01/


  • 1
    $\begingroup$ This is not a list of definitions. This is a list of sums. For example, if you open the link concerning the Hermite function, you will see 16 series. The first one is $$\sum_{n=0}^\infty \frac{H_n(z)w^n}{n!}=e^{2zw-w^2}$$ The series definitions of the special functions are not in the above list. Of course, it is easy to find them on the Wolfram's functions site. $\endgroup$ – JJacquelin Sep 11 '17 at 20:03
  • $\begingroup$ Yes, that's what I meant with includes a lot of formulas, that's good (!) to show the OP, that there are endless many examples. But e.g. the link of BernoulliB contents only a possible definition. :-) The OP doesn't say anything more about what he likes to know but I speculate that he expects some exotic series which cannot be find easily. $\endgroup$ – user90369 Sep 12 '17 at 5:19
  • $\begingroup$ OK. From the incomplete list of references, one can extract some series very difficult to find by ourselves as it is probably expected by the OP. I have not carried out a screening : it should be a tiresome task, not useful until the question be less wide. $\endgroup$ – JJacquelin Sep 12 '17 at 6:55
  • $\begingroup$ I agree. Till the OP doesn't specify more we have endless many possibilities for an answer. :-) $\endgroup$ – user90369 Sep 12 '17 at 9:44
  • $\begingroup$ Thus, it is up to the OP to chose among the references and examples provided in the answers what is the most useful for his purpose. I think that it is not possible for us to help him more. $\endgroup$ – JJacquelin Sep 12 '17 at 10:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.